an alternative means of spectroscopic...
TRANSCRIPT
AN ALTERNATIVE MEANS OF SPECTROSCOPIC IMAGING FOR SUPERFICIAL WOUND HEALING
PROCESS MONITORING
SHEENA PUNAI ANAK PHILIMON
UNIVERSITI TUN HUSSEIN ONN MALAYSIA
AN ALTERNATIVE MEANS OF SPECTROSCOPIC IMAGING FOR
SUPERFICIAL WOUND HEALING PROCESS MONITORING
SHEENA PUNAI ANAK PHILIMON
A thesis submitted in
fulfillment of the requirement for the award of the
Degree of Master of Electrical Engineering
Faculty of Electrical and Electronic Engineering
Universiti Tun Hussein Onn Malaysia
AUGUST 2016
iii
To my beloved parents and supervisor, thank you.
iv
ACKNOWLEDGEMENT
Foremost, I would like to express my heartfelt gratitude to the Almighty God for the
wisdom and perseverance that He has bestowed upon me during this research, and
indeed, throughout my life.
My deepest appreciation goes to my supervisor, Dr. Audrey Huong for her
unwavering guidance, enthusiastic encouragement and mentorship in all the time of
research and writing of this thesis. I could not have imagined having a better
supervisor. I acknowledge, with appreciation, my debt of thanks to Dr. Xavier Ngu,
for offering me the convenience to approach all facilities available in the laboratory.
I would also like to acknowledge the medical staffs of Pusat Kesihatan Universiti,
UTHM and all volunteers involved for their worthy support and cooperation
throughout the course of this research.
I extend my thanks to all my friends especially Qing Shi, Feng Ng and
everyone who have, directly or indirectly, helped me beyond their abilities. Finally,
special thanks to my beloved parents and sister, for their unceasing moral support
and precious love, who have always stood by me through so many hard times.
v
ABSTRACT
To date, oximetry is considered as an ideal approach to monitor one's
microcirculation. Quantitative information of wound tissue oxygenation is beneficial
for proper wound care in response to different treatment. The conventional method of
oximetry using a fingertip pulse oximeter is limited to certain body parts and
required contact, hence, is deemed unsuitable for wound assessment. This research
describes a non-invasive and non-contact multispectral spectroscopy approach for
optical monitoring of changes in oxyhemoglobin saturation level during wound
healing process. Non-contact reflectance data captured from the wounded skin site
using a monochromatic imaging system in the wavelength range of 520−600 nm are
mathematically analyzed and fitted using Extended Modified Lambert Beer model to
give the best estimation of percent transcutaneous oxygen saturation, StO2.
Experimental works conducted on ten Asian subjects with different wound sites
revealed progressive changes in StO2 level throughout the healing process. The
results revealed a significant increase in StO2 that reaches its peak in between fourth
and sixth day of wound healing. Quantitative analysis of wound oxygenation level
using line profiler based on StO2 mapping shows high StO2 values of greater than
90% in wounded skin sites while StO2 in the range of 30−50% is observed along the
adjacent unwounded skin region. The results from this study show feasibility of
using this technique to provide visual progression of wound via tissue oxygenation
status, suggesting the prospective implementation of this system in hospitals as an
alternative means to assess wound healing.
vi
ABSTRAK
Sehingga kini, oximetry dianggap sebagai satu pendekatan ideal untuk memantau
peredaran mikro darah. Maklumat kuantitatif pengoksigenasi tisu luka bermanfaat
untuk penjagaan luka yang betul sebagai tindak balas kepada rawatan yang berbeza.
Kaedah umum menggunakan nadi oximeter hujung jari adalah terhad kepada
bahagian-bahagian badan tertentu dan melibatkan penyentuhan, oleh itu, dianggap
tidak sesuai untuk penilaian luka. Kajian ini menerangkan pendekatan bukan invasif
dan tidak bersentuhan multispectral spectroscopy untuk pemantauan optik perubahan
tahap tepu oxyhemoglobin semasa proses penyembuhan luka. Data pantulan yang
diambil dari permukaan kulit luka menggunakan sistem pengimejan monokromatik
dalam julat panjang gelombang antara 520−600 nm dianalisis secara matematik
menggunakan Extended Modified Lambert Beer model untuk memberi anggaran
terbaik peratus ketepuan oksigen transcutaneous, StO2. Kajian eksperimen yang
dijalankan ke atas sepuluh subjek Asia dengan luka yang berbeza mendedahkan
perubahan progresif tahap StO2 sepanjang proses penyembuhan. Hasil kajian
menunjukkan peningkatan yang ketara tahap tahap StO2 yang mencapai nilai puncak
antara hari keempat dan hari keenam penyembuhan luka. Analisis kuantitatif tahap
pengoksigenan luka menggunakan profil garis berdasarkan pemetaan StO2
menunjukkan peningkatan dalam nilai StO2 melebihi 90% dalam kulit luka manakala
StO2 dalam julat 30−50% diperhatikan sepanjang kawasan kulit yang tidak luka.
Hasil kajian ini menunjukkan kesesuaian teknik ini dalam menyediakan
perkembangan visual luka melalui status pengoksigenasi tisu dan dengan itu
mencadangkan pelaksanaan sistem ini di hospital sebagai alternatif untuk menilai
penyembuhan luka.
vii
LIST OF ASSOCIATED PUBLICATIONS
Journals
1. A. K. C. Huong, S. P. Philimon, and X. T. I. Ngu, "Non-invasive estimation of
blood oxyhemoglobin and carboxyhemoglobin saturations using cumulant based
forward model," ARPN Journal of Engineering and Applied Sciences, vol. 10, pp.
8421-8426, 2015.
2. S. P. Philimon, A. K. C. Huong, and X. T. I. Ngu, "Multispectral imaging system
for quantitative assessment of transcutaneous blood oxygen saturation," Jurnal
Teknologi, vol. 77, 2015.
3. S. P. Philimon, A. K. C. Huong, and X. T. I. Ngu, "Investigation of multispectral
imaging technique for optical monitoring of mean blood oxygen saturation,"
ARPN Journal of Engineering and Applied Sciences, vol. 11, pp. 3951-3956,
2016.
Conference proceedings
1. A. K. C. Huong, S. P. Philimon, and X. T. I. Ngu, "Noninvasive monitoring of
temporal variation in transcutaneous oxygen saturation for clinical assessment of
skin microcirculatory activity," in International Conference for Innovation in
Biomedical Engineering and Life Sciences, 2016, pp. 248-251. Springer
Singapore.
Poster conference
1. "Reflectance spectroscopy system for noninvasive diagnosis of carbon monoxide
poisoning." FKEE Postgraduate Poster Conference, Hari Transformasi Minda
(FKEE, UTHM), 21 September 2015.
viii
TABLE OF CONTENTS
TITLE i
DECLARATION ii
DEDICATION iii
ACKNOWLEDGEMENT iv
ABSTRACT v
LIST OF ASSOCIATED PUBLICATIONS vii
CONTENTS viii
LIST OF TABLES xii
LIST OF FIGURES xiii
LIST OF SYMBOLS AND ABBREVIATIONS xvii
LIST OF APPENDICES xviii
CHAPTER 1 INTRODUCTION 1
1.1 Background of study 1
1.2 Problem statement 3
1.3 Aim 4
1.4 Objectives 4
1.5 Scopes of study 5
1.6 Outline of the thesis 6
ix
CHAPTER 2 LITERATURE REVIEW 7
2.1 Visible light optical spectroscopy 7
2.1.1 Multispectral imaging techniques 11
2.2 Optical properties of the human skin model 12
2.3 Quantitative analysis of percent mean blood
oxygen saturation
16
2.3.1 Kubelka Munk theory 17
2.3.2 Cumulant based attenuation model 17
2.3.3 Cubic function 18
2.3.4 Power law model 19
2.3.5 Lambert − Beer law 19
2.3.6 Modified Lambert Beer law 20
2.3.7 Extended Modified Lambert Beer model 21
2.4 Wound healing assessment 25
2.4.1 Categories of wounds 25
2.4.2 Wound healing stages 26
2.4.3 Oxygen requirement in wounded skin
tissue
28
2.4.4 Quantitative wound assessment 29
2.5 Summary 32
CHAPTER 3 RESEARCH METHODOLOGY 33
3.1 Multispectral imaging system 33
3.1.1 Experimental instrument and optical
system
34
3.1.2 Performance of multispectral imaging
system
37
3.1.3 Multispectral image correction and
image acquisition
38
3.1.4 Analytical technique validation using
multispectral data
39
3.2 Experimental subjects and procedure 40
x
3.2.1 Data acquisition and data cube
formation
41
3.3 Extended Modified Lambert Beer model and
non-linear fitting algorithm
42
3.3.1 Iterative fitting procedure 42
3.4 Collection of StO2 line profile 45
3.4.1 Selection criteria for experimental
subjects
45
3.4.2 Preprocessing of multispectral data 46
3.5 Summary 49
CHAPTER 4 RESULTS AND DISCUSSION 50
4.1 Results from multispectral imaging system 50
4.1.1 Performance of multispectral imaging
system
50
4.1.2 Preliminary results on transcutaneous
oxygen saturation measurement
52
4.2 Non-invasive assessment of wounded skin
samples
57
4.2.1 Wound subject A 57
4.2.2 Wound subject B 59
4.2.3 Wound subject C 61
4.2.4 Wound subject D 63
4.3 Quantitative analysis of wound tissue
oxygenation
65
4.3.1 Prediction of transcutaneous blood
oxygen saturation for subject A
65
4.3.2 Prediction of transcutaneous blood
oxygen saturation for subject B
68
4.3.3 Prediction of transcutaneous blood
oxygen saturation for subject C
71
4.3.4 Prediction of transcutaneous blood
oxygen saturation for subject D
74
xi
4.4 Comparison of oxygen saturation in wound and
unwounded skin using StO2 line profiler
80
4.4.1 Wound subject E 80
4.4.2 Wound subject F 83
4.4.3 Wound subject G 85
4.4.4 Wound subject H 87
4.4.5 Wound subject I 89
4.4.6 Wound subject J 91
4.5 Summary 95
CHAPTER 5 CONCLUSION 96
5.1 Conclusion 96
5.2 Research contribution 99
5.3 Recommendation for future work 100
REFERENCES 101
APPENDIX 108
VITA 112
xii
LIST OF TABLES
2.1 Different skin layers and their thickness used by
Meglinski and Matcher [39]
14
2.2 Percent mean blood oxyhemoglobin, SmO2 and
carboxyhemoglobin, SmCO saturation in smoking and
non-smoking individuals as reported by Huong and
Ngu [56]
24
4.1 Estimated percent mean StO2 for volunteers (referred
to as volunteer A − D) at rest and during blood flow
occlusion using EMLB model
53
4.2 A comparison of percent StO2 for volunteers at rest and
during arterial blood occlusion experiment obtained
from this work and that reported in previous literatures
55
4.3 Quantitative percent StO2 results based on processed
multispectral images of wound subject A
65
4.4 Quantitative percent StO2 results based on processed
multispectral images of wound subject B
68
4.5 Quantitative percent StO2 results based on processed
multispectral images of wound subject C
71
4.6 Quantitative percent StO2 results based on processed
multispectral images of wound subject D
74
xiii
LIST OF FIGURES
2.1 The basic operating principle of a fingertip pulse
oximeter
7
2.2 Optical path of transmitted visible light across
different skin layers [17]
8
2.3 General model of continuous light in a semi
infinite medium
9
2.4 Isosbestic wavelengths identified at 546 nm and
569 nm when the extinction coefficients of HbO2
and Hb are the same
10
2.5 The multilayered anatomical structure of the
human skin tissue [38]
13
2.6 Absorption coefficient of HbO2, Hb, water and
melanin compiled by Meglinski and Matcher
[39]
15
2.7 The wavelength dependent molar extinction
coefficients of oxyhemoglobin (2HbOε ),
deoxyhemoglobin ( Hbε ) and carboxyhemoglobin
( COHbε ) compiled from the reports of Zijlstra et
al. [46]
23
2.8 The multifaceted stages involved in
physiological wound healing process [1]
28
2.9 Variability in the oxygen saturation line profiles
of diabetic related ulcers [11]
30
2.10 Changes in oxyhemoglobin level after wound
surgery is performed on Day 0 [10]
31
xiv
3.1 Schematic diagram of the multispectral imaging
system experiment setup
33
3.2 Optical arrangement of the multispectral imaging
system
35
3.3 Optical light path through the plano-convex lens 36
3.4 Optical arrangement of dispersive elements
inside the monochromator
37
3.5 Validation of the monochromator wavelength
using a spectrometer
38
3.6 An example of selected wound region captured
for multispectral data measurement
41
3.7 Unconstrained non-linear iterative fitting
procedure using MATLAB fminsearch function
44
3.8 The targeted skin region (indicated by the red
box) captured with the CCD camera showing a
segment of the wounded and its adjacent
unwounded skin
46
3.9 The multispectral image stack used in forming
the data cube
48
4.1 Graph showing light spectrum at centre
wavelength of 550 nm with FWHM of 2.6 nm as
detected by spectrometer
51
4.2 An example of multispectral reflectance data at
wavelength 520 nm (monochrome image)
collected from the right index finger of human
subject
52
4.3 Multispectral images taken at interval
wavelengths aligned as a stack
53
4.4 The multispectral image of right index finger of
volunteer A
54
4.5 Images of wound healing progress of subject A
upon receiving treatment
58
xv
4.6 Initial condition of the wounded foot following
accident
59
4.7 Images of wound healing progress of subject B
upon receiving treatment
60
4.8 Images of wound healing progress of subject C
upon receiving treatment
62
4.9 Images of wound healing progress of subject D
upon receiving treatment
64
4.10 Quantitative changes of percent mean blood
transcutaneous oxygen saturation, StO2, in
wounded skin tissue during wound healing
assessment (subject A)
66
4.11 The StO2 maps for reflectance data measured
from the wound of subject A
67
4.12 Quantitative changes of percent mean blood
transcutaneous oxygen saturation, StO2, in
wounded skin tissue during wound healing
assessment (subject B)
69
4.13 The StO2 maps for reflectance data measured
from the wound of subject B
70
4.14 Quantitative changes of percent mean blood
transcutaneous oxygen saturation, StO2, in
wounded skin tissue during wound healing
assessment (subject C)
72
4.15 The StO2 maps for reflectance data measured
from the wound of subject C
73
4.16 Quantitative changes of percent mean blood
transcutaneous oxygen saturation, StO2, in
wounded skin tissue during wound healing
assessment (subject D)
75
4.17 The StO2 maps for reflectance data measured
from the wound of subject D
75
xvi
4.18 The graph of compiled percent mean StO2 versus
relative number of days estimated for the wounds
of four recruited subjects
77
4.19 (Left) The StO2 profile of wound subject E
measured across spatial axis y: 60, (Right) The
selected wound region
82
4.20 (Left) The StO2 profile of wound subject F
measured across spatial axis y: 60, (Right) The
selected wound region
84
4.21 (Left) The StO2 profile of wound subject G
measured across spatial axis y: 80, (Right) The
selected wound region
86
4.22 (Left) The StO2 profile of wound subject H
measured across spatial axis y: 20, (Right) The
selected wound region
88
4.23 (Left) The StO2 profile of wound subject I
measured across spatial axis y: 140, (Right) The
selected wound region
90
4.24 (Left) The StO2 profile of wound subject J
measured across spatial axis y: 100, (Right) The
selected wound region
92
xvii
LIST OF SYMBOLS AND ABBREVIATIONS
A Light attenuation, λ
C Concentration of absorber, mol L-1
CCD Charge-coupled detector
CO Carbon monoxide
COHb Carboxyhemoglobin
d Light pathlength
EMLB Extended Modified Lambert Beer
fL Focal length, mm
G0 Light attenuation offset
G1 Absorption dependent light attenuation
Hb Deoxygenated hemoglobin
HbO2 Oxyhemoglobin
I Light intensity
SaO2 Arterial blood oxygen saturation, %
SmCO Mean blood carboxyhemoglobin saturation, %
SmO2 Mean blood oxyhemoglobin saturation, %
SO2 Blood oxygen saturation, %
StO2 Transcutaneous oxygen saturation, %
T Total absorbers concentration, mol L-1
ε Molar extinction coefficient, mm-1 M-1
Δε Absorptivity difference, mm-1 M-1
λ Wavelength, nm
µa Absorption coefficient, mm-1
µs Scattering coefficient, mm-1
ρO2 Partial pressure of oxygen, mmHg
Ø Diameter, mm
xviii
LIST OF APPENDICES
APPENDIX TITLE PAGE
A Informed consent statement 108
B Initial wound images of subject A and B 110
C Initial wound images of subject C and D 111
D "Non-invasive estimation of blood oxyhemoglobin
and carboxyhemoglobin saturations using cumulant
based forward model," ARPN Journal of
Engineering and Applied Sciences, vol. 10, pp.
8421-8426, 2015.
112
E "Multispectral imaging system for quantitative
assessment of transcutaneous blood oxygen
saturation," Jurnal Teknologi, vol. 77, 2015.
113
F "Investigation of multispectral imaging technique for
optical monitoring of mean blood oxygen
saturation," ARPN Journal of Engineering and
Applied Sciences, vol. 11, pp. 3951-3956, 2016.
114
G "Noninvasive monitoring of temporal variation in
transcutaneous oxygen saturation for clinical
assessment of skin microcirculatory activity," in
International Conference for Innovation in
Biomedical Engineering and Life Sciences, 2016, pp.
248-251. Springer Singapore.
115
CHAPTER 1
INTRODUCTION
1.1 Background of study
In cases of chronic wound healing, wound is known to take several months to
heal unlike normal wound healing process. In isolated cases the wound may not be
able to regain its anatomical and functional structure which may end up with
amputation of the concerned body part. According to a report in 2009, approximately
eight million people suffered from chronic wounds in Europe [1]. A retrospective
study by the European Wound Association Management (EWMA) in 2010 has
reported at least 1−1.5% of the industrialized world's population will be diagnosed
with problematic wound at any single time [2]. Likewise in Malaysia, statistical
study on diabetes mellitus (DM) has shown a drastic increase from 0.6% in 1960 to
2.1% in 1982, 6.3% in 1986, 8.3% in 1996, 14% in 1998 and between 16−18% as of
2005 [3-5]. Based on these figures, about 20% of the population suffered from
diabetic foot infections which are at risk of amputation [5]. The prevalence of non-
healing chronic wounds is increasingly alarming and this has affected the lifestyle of
many people both emotionally and financially.
In view of this, proper assessment of wound based on analysis of pathological
factors is necessary before the condition of the wound deteriorates. One of the key
determinants for assessing wound healing is by analyzing the fractional
concentration of blood hemoglobin level which corresponds with transcutaneous
blood oxygen saturation level (StO2). Having detailed information of the wound
condition will provide an informational guideline for clinical practitioners to identify
2
detrimental factors that delay wound healing, such as wound hypoxia [1]. Proper
diagnosis and treatment of StO2 at the surrounding wound tissue will ensure
progressive improvement in healing of wounds.
Meanwhile, oxygen has been widely addressed for its imperative role
throughout the process of wound healing. Proper oxygen perfusion is necessary in
order to establish recovery and full functionality of the anatomical skin structure.
Oxygen is regarded as an essential component in bacterial killing by neutrophils and
collagen production by fibroblast [6, 7]. During the various stages of wound healing,
a substantial demand of oxygen is required by the wound and the surrounding tissue
area for optimal granulation of new tissue cells [8]. Acute wounds are normally
healed within an orderly and timely period after undergoing the essential healing
stages following the injury. These stages are sequential starting with the
inflammatory stage, proliferative stage that includes reepithelialization of tissue
granulation, and finally remodeling of tissue structure [1, 9].
For years, the use and application of optical reflectance spectroscopy are
much discussed subjects in biomedical research. This is owing to the simplicity of
the system and its non-invasive nature. Many of these researches focused on using
the corresponding system to characterize and determine optical properties of
biological tissues such as skin wound site [10, 11]. An applicable diagnostic tool that
could provide in-vivo measurement of StO2 and present these data using a
comprehensible map would be of high benefit to clinical practitioners. This would
provide a visual insight of the wound condition to target on the specific problematic
wound site. Due to this reason, wound tissue oximetry is considered as one of the
suitable parameters to evaluate wound healing outcome.
3
1.2 Problem statement
A reliable and accurate method of providing quantitative measurement of wound
tissue oxygenation level is highly sought after. The conventional way of assessing
wound is simply based on the physical appearance such as surface area of the wound,
wound color and odor whereby other underlying pathological factors are often
overlooked [9]. This is not accurate as a more holistic approach is required for
wound assessment rather than just assessing the surface appearance of the wound.
Besides, several underlying symptoms involving cases of non-healing chronic
wounds such as extreme near-anoxic hypoxia are sometimes unidentified which
resulted in a delay of medical decisions due to unguided improper therapeutic
approach [12]. This is both time consuming and an economic burden to many due to
the high cost of medical care and treatment when instead early decisions could be
made through early identification of healing outcome based on measurement of
wound oxygenation status. Therefore, a non-invasive and non-contact means of
blood oxygen saturation (SO2) measurement for this application has gained
increasing interest in the medical arena. This is in accordance to the major drawbacks
of pulse oximeter, which clinical measurement means is via the use of a finger clip
for measurement of arterial blood oxygen saturation (SaO2), rendering it unsuitable to
monitor wounded skin grafts. In addition, high accuracy of a pulse oximeter is
limited to SaO2 value of greater than 70% which restricts the application of this
device among anemic patients [13].
4
1.3 Aim
The aim of this research is to demonstrate the use of multispectral imaging system
for skin oximetry and to investigate changes in oxygen consumption at local skin
tissues during wound healing.
1.4 Objectives
The following objectives were carried out to achieve the aim of this research:
a) To develop a non-invasive technique for continuous monitoring and
measurement of wound oximetry.
b) To investigate the relationship between wound healing rate and tissue
oxygenation level.
c) To compare transcutaneous oxygen saturation value between the wound site
and adjacent skin tissue.
5
1.5 Scopes of study
The scopes of this research are:
a) To design and develop a multispectral imaging system for non-invasive
measurement of skin oximetry.
A monochromator and a charge-coupled detector (CCD) camera are
employed in the experimental system to acquire the reflectance spectra of the
skin.
b) To employ an attenuation model to deduce one's transcutaneous blood
oxygen saturation via a fitting algorithm.
This work employed the Extended Modified Lambert Beer model developed
by Huong and Ngu [16] for quantification analysis of acquired data.
c) To determine factors (skin sites and life styles) that would affect the
differences in the estimated StO2.
This technique is demonstrated on different wound sites of human subjects to
investigate differences in the estimated StO2 with wound sites and their life
styles (smokers/non-smokers).
6
1.6 Outline of the thesis
In this thesis, a non-invasive multispectral imaging spectroscopy is developed for
quantitative analysis of wound tissue oxygenation using EMLB model. The
quantification technique employed in this work is based on the extinction coefficient
of hemoglobin derivatives within a wavelength range of 520−600 nm to give the best
estimation of StO2 fractional concentration. The rest of this thesis is organized as
follows:
I. Introduction
a. Background of study and research motivation
b. Aim and objectives
c. Scopes of study
II. Literature review
a. Optical reflectance spectroscopy techniques
b. Optical properties of the human skin model
c. Analytical models for quantitative analysis of percent mean blood
oxygen saturation
d. A review of the role of oxygen in wound healing
III. Research methodology
a. Development of the multispectral imaging system
b. Calibration of spectroscopy system and multispectral image correction
c. Analysis of multispectral data using Extended Modified Lambert Beer
model and non-linear fitting algorithm
d. Data collection of experimental wound subjects
IV. Results and discussion
a. Preliminary results for analytical technique validation using
multispectral data
b. Quantitative analysis of wound tissue oxygenation during healing
c. Comparison of oxygen saturation level in wound and unwounded skin
sites using StO2 line profiler
V. Conclusion and future recommendations
CHAPTER 2
LITERATURE REVIEW
2.1 Visible light optical spectroscopy
The common way of measuring blood oxygen saturation is via means of a fingertip
pulse oximeter. The operating principle of pulse oximeter is based on the ratio of
detected intensity of two wavelengths, i.e. the red wavelength (660 nm) and infrared
wavelength (880 nm) [12]. This device is made up of two light emitting diodes
(LEDs) and a photodetector as shown in Figure 2.1. Light transmitted through the
finger is detected by a photodetector. If the oxygenated hemoglobin concentration is
high, absorption of infrared light would be high, followed by poor absorption of red
light. Otherwise is true for high deoxygenated hemoglobin (Hb) concentration.
Figure 2.1: The basic operating principle of a fingertip pulse oximeter
8
Continuous measurement and quantification of tissue oxygen saturation can
be achieved via a non-invasive and non-contact optical approach known as optical
reflectance spectroscopy. Spectroscopy imaging is a technique commonly practiced
nowadays in medical diagnosis and prognosis to provide intensity information of a
sample across a selected wavelength range. Spectral measurement from both the skin
surface and below the skin surface is obtained when light is sufficiently reflected,
transmitted, and absorbed by the skin and tissues. Light in the visible wavelength
range that irradiates the skin surface will pass through different layers, starting with
the stratum corneum, epidermis layer and the dermis layer of the human skin as
illustrated in Figure 2.2. Although a minute fraction of light will be reflected back
due to the changes in refractive index between air and stratum corneum, the
remaining transmitted light such as the red light (e.g. at wavelength of λ = 600 nm)
could infiltrate the dermis layer up to a depth of 500 nm [17].
Figure 2.2: Optical path of transmitted visible light across different skin layers [17]
An advantage of the optical spectroscopic technique is its non-invasive
attribute that complemented the technique as an ideal wound oximeter. Different
techniques have been developed to provide detailed information regarding the
composition of oxygen level beneath the human skin based on the obtained
9
spectroscopic data. Amongst these techniques are the use of a library of data
simulated using Monte Carlo method or diffusion model [18], Extended Modified
Lambert Beer (EMLB) model [16] and a non-linear fitting model [19] to fit to the
measurement data.
Theoretically, the basic of this technique is based on diffuse reflectance on
tissue at visible wavelength range to determine the fractional concentration of
absorbers [20, 21]. A simple vindication on the basic concept of spectroscopic
technique is shown in Figure 2.3 using the model given by Wu Ri et. al [22]. This
model consisted of a light source and detector place upon the tissue surface to
examine the diffusion theory of photon path in a semi infinite medium based on the
distance between the source and detector, ρ . In this model, HbO2, Hb and water are
primary absorbers assumed to be present within the tissue. Wu Ri et al. [22]
explained that oxygen concentration in tissue can be mathematically resolved based
on the knowledge of absorbers' absorption and scattering coefficients.
Figure 2.3: General model of continuous light in a semi infinite medium
Taking into account the molecular properties of human tissue chromophores
as absorbing medium, the penetration of light within the visible wavelength range is
calibrated to determine the isosbestic absorption spectra which can be consequently
utilized to measure total tissue hemoglobin concentration [23]. The isosbestic point,
identified at 546 nm and 569 nm in Figure 2.4, is referred to as the intersection point
10
at which the extinction coefficient of HbO2 and Hb is known to have the same value
[11, 20]. The isosbestic point is typically consigned as a correction point during
calculation of StO2 because the absorptivity of hemoglobin chromophores of the
measured attenuation at the equivalent wavelength is invariant with changes in StO2
[24, 25]. Meanwhile, in the work of Duling and Pittman [26] a non-isosbestic
wavelength pair is used to solve the StO2 and an isosbectic wavelength pair is used to
define a straight line because of the near proximity between these wavelength pairs
in the graph of attenuation, A, versus absorption coefficient, aµ , which renders a
more precise representation of the Modified Lambert Beer (MLB) law equation [27,
28]. In optical spectroscopic approach, the photon propagation pathlength
corresponded linearly with the light attenuation in a non-scattering medium.
However, the photon pathlength varies with light wavelength in the cases of
scattering medium. This renders an erroneous estimation of the optical properties
using the corresponding assumption, thus evaluation of a whole spectrum is opted for
[29].
Figure 2.4: Isosbestic wavelengths identified at 546 nm and 569 nm when the extinction coefficients of HbO2 and Hb are the same
Isosbestic point
11
2.1.1 Multispectral imaging techniques
Previous works on spectroscopic analysis of skin microcirculation can be categorized
into several techniques namely optical point spectroscopy [16] and imaging
modalities such as multispectral imaging [18, 30] and hyperspectral imaging [11,
27]. Compared to point spectroscopy technique, the multispectral imaging technique
was has an advantage over the previous considering the ability of the CCD camera to
detect the intensity of light spectrum at wider image coverage as opposed to the
preceding technique which assessment is limited to a single point. These limitations
are a frequent concern when using the pulse oximeter for clinical diagnosis, of which
the concept is similar to a point probe spectrometer by detecting the emitted spectral
value at one point. Among the limitations related to these subjects are such as
inadequate blood volume for analysis, irregular pulsation, exposure of light per unit
area and variation in skin color [24]. Furthermore, a constant distance between the
detector and the source that is suitable with the absorption and scattering coefficient
of medium is vital throughout the spectroscopic measurement. In a CCD camera, the
detector is integrated inside the camera and the main apprehension is not the distance
between source and detector but to analyze the pixels of the multispectral images as
these distances vary among pixels [22, 31].
Meanwhile, a comparison between hyperspectral imaging system and
multispectral imaging system is usually made in view of the number of spectral
bands. Contrary to the concept of a hyperspectral system that deals with hundreds to
thousands of spectral bands, multispectral system captured images at a relatively
discrete and narrow band, approximately not more than 20 bands in each pixel
produced [32]. Due to this reason, a longer computational time is required during the
analysis of hyperspectral data sets as compared to the latter. These differences are
also apparent in terms of comparison between the imaging sensors. In a hyperspectral
imaging system, a single sensor is used to capture a few spectral bands over a set of
proximate wavelength range [33]. This is contrary to a multispectral system that
requires a multiple number of sensors to capture a set of spectral bands which are not
characteristically proximate. In multispectral imaging system, a CCD camera
produced distinguished spectroscopy images at multiple wavelengths obtained from
the skin site of interest. These multiwavelength images are aligned accordingly to
12
form a multispectral data cube which could provide spatial and spectral information
[33, 34].
In the work by M. Aikio [35], the author has explained the concept of both
spatial and spectral wavelengths simultaneously detected by the CCD which is then
produced as a stack of images and specifically dubbed it as a data cube. Applying
this concept into recent microcirculatory related work, Wu-Ri et al. [22] expressed
the multilayered images as a form of three dimensional hypercube, given by the
absorption coefficient, scattering coefficient and source-detector separation distance,
hence, analyzing the data cube as a representation of light attenuation at
corresponding wavelengths. The approach that adopts multispectral images to
evaluate the local changes in HbO2 and Hb could possibly provide a reliable
assessment of wound healing and skin grafts. Additionally, the images could
potentially provide quantitative and detailed visual information in assessing vascular
tumor up to two to three millimeters deep within the skin layer without any physical
contact [18, 36].
2.2 Optical properties of the human skin model
Human skin is composed of a complex multilayered tissue structure, with each tissue
structure incorporating unique optical properties which are taxing to determine. The
top most layer of the skin structure is made up of the stratum corneum and the dermis
layer itself is made up of multiple layers of structures as illustrated in Figure 2.5.
Amongst these layers, collagen, fibroblasts, blood vessels and sensory receptors are
found in abundance within the dermal layer, which is primarily made up of
mesoderm [37].
13
Figure 2.5: The multilayered anatomical structure of the human skin tissue [38]
Meglinski and Matcher [39] employed the Monte Carlo simulation by
modeling the skin into seven layers with respective thickness as tabulated in Table
2.1, taking into consideration the absorption spectra and optical properties of each
layer. Later, Zhang et al. [40] simplified this seven-layered model to a three-layered
model consisting of epidermis, papillary upper dermis and reticular dermis to
determine the effects of optical properties in each layer on the measured reflectance
spectrum. A study by Barun et al. [41] on the optical properties of human skin
structure using the three-layered skin model that consists of the stratum corneum,
epidermis and dermis layer shows a high correlation between the irradiated spectral
wavelength and the penetration depth of light into tissue. The corresponding author
employed a laser light source in the wavelength range of 441−1062 nm to estimate
the penetration depth of light into tissue of human subjects with normal skin and
pathologically altered skin. Subjects with dipegmented skin disorder were observed
to have deeper penetration depth as compared to normal skin because of the thinner
epidermis layer of 60 µm in the former which allows light distribution to be assessed
mainly in the dermis layer.
14
Table 2.1: Different skin layers and their thickness used by Meglinski and Matcher [39]
Layer Thickness (µm)
Stratum corneum 20 Epidermis 80 Dermis
• Papillary dermis • Upper blood dermis • Reticular dermis • Deep blood net dermis
150 80
1500 100
Subcutaneous fat 6000
Inherently, the computational skin modeling clearly illustrated the thickness
of each layer to determine the spatial distribution of absorbers within the skin tissue.
Melanin and hemoglobin derivatives (HbO2 and Hb) are most significant in the skin
epidermis and dermal layers in the visible wavelength range of 500 nm to 620 nm,
respectively [10, 40, 42]. However, melanin is frequently disregarded as no
distinguished changes were observed in melanin variation within this wavelength
range [27, 42, 43]. In the study of blood oxygen saturation, the concentration of
melanin is irrelevant but for other studies related to the investigation of skin
pigmentation or jaundice the absorptivity of melanin is often taken into consideration
[44]. Although other absorbers such as water, bilirubin and other tissue
chromophores are present under the skin, these absorbers are less apparent as
compared to the previously stated derivatives [45]. The absorption coefficients of the
stated absorbers are assessed in the wavelength range of 400−1100 nm as presented
in Figure 2.6. It must be noted however, the validity of the spectral absorption
coefficient data shown is only applicable when assessing hemoglobin derivatives in
adult human. This is due to the reason that although both data will produce a similar
spectral graph curve, there is a diminutive variation of hemoglobin absorptivity
between human adult and fetal which must not be disregarded during analysis of
mean StO2 [46].
15
Figure 2.6: Absorption coefficient of HbO2, Hb, water and melanin compiled by Meglinski and Matcher [39]
Based on Figure 2.6, absorbers such as HbO2, Hb, melanin water and other
chromophores present in the blood medium is assigned to a unique absorption
coefficient value at different wavelength. These values are used as a reference to
assign the attenuation spectrum with the corresponding wavelength dependent
absorption coefficient by means of a stochastic quantitative model. The given data
shows the absorbers are analyzed at a visible and near-infrared (NIR) wavelength,
but simulations by Meglinski and Matcher [39] revealed an analogous value
comparable with experimental results reported when using visible light in the
wavelength range of 450−600 nm instead of the NIR spectral wavelength range
between 700−1100 nm. A main factor leading to this is the optical changes in
scattering of skin tissue which gradually reduced as the wavelength is varied from a
visible to a NIR wavelength range [24, 39].
16
2.3 Quantitative analysis of percent mean blood oxygen saturation
An individual's transcutaneous blood oxygen saturation within the tissue layer can be
quantitatively determined based on several techniques. These techniques include the
Kubelka Munk theory [17, 29, 47], cumulant based attenuation model (CM) [27], the
linear MLB law [11, 26] and non-linear fitting models such as the Extended
Modified Lambert Beer model [16], cubic function [42] and the power law model
[18]. The analytical equation is normally derived based on the Monte Carlo
simulation or diffusion approximation [48].
In the work by Pifferi et al. [19], the photon migration in a continuous wave
(CW) measurement is explicated using the Monte Carlo simulation, whereby the
values of absorption coefficient, aµ , and scattering coefficient, sµ , are extracted by
referring to a look up table. Referring to this work [19], the accuracy of the fitted
model is evaluated in terms of relative error, ε given by Equation 2.1:
f e
e
µ µε
µ−
= (2.1)
This equation, defined by parameters eµ and fµ , representing the effective
value of aµ and sµ of the characterized medium and the corresponding fitted value,
respectively. This model produced a relative error <10% evaluated using Monte
Carlo model. This is contrary to the result obtained via diffusion approximation,
which provided a relative error >30% for both optical coefficients considered, thus
diffusion method is unreliable.
17
2.3.1 Kubelka Munk theory
The Kubelka Munk theory delineates the propagation of a perfectly, diffuse
irradiation through a one-dimensional isotropic slab in which regular reflection at the
boundary is neglected [17, 49]. The propagation of light is expressed as a complex
relation of light transmission and reflection approximated using the equation written
in the form of [27],
( ) ( ) ( )( )
a s1r
s
2coshy
µ λ µ λλ
µ λ− +
=
(2.2)
where ( )ry λ is given as the negative natural logarithm of the measured reflectance
spectrum. The symbols aµ and sµ in this equation represent the absorption
coefficient and scattering coefficient in the dermis layer, respectively.
Although Kubelka Munk theory is simple for quantitative assumption of skin
optics, there are drawbacks to this theoretical model wherein the accuracy of the
experimental measurement is a subject of dispute. These drawbacks are reflected by
several aspects such as the measurement which are solely based on the conjecture of
isotropic scattering of photons in a medium and mismatched boundary condition of
the irradiation source. In fact, it is almost impossible to accomplish a uniformly
diffused or isotropic radiation in practice due to varying tissue structures and
measurement parameters. An excellent example of experimental work using the
Kubelka Munk theory is demonstrated by Caspary et al. [29], using the
corresponding analytical model to investigate temporal variation of oxyhemoglobin
saturation in human skin tissue based on scattering properties of light spectra.
2.3.2 Cumulant based attenuation model
Cumulant based attenuation model is previously used in the work of Huong [27] for
an improved estimation of percent StO2. A comprehensive development and
derivation of the model has been extensively explained by the author in the
corresponding work, and it is briefly summarized here again. The non-logarithm
relationship between temporal point spread function (TPSF) cumulants and
18
wavelength in a non-absorbing infinite slab is substituted into the moment dependent
intensity model proposed by Sassaroli et al. [50], to give an expression of light
attenuation represented as a summation of cumulants as follows:
( ) ( )( ) ( )( ) ( )0 0 a aexp exp aA a b f g h k mλ λ µ λ λ µ λ µ λ= + + − − − + (2.3)
where. 0a , 0b , f , g , h , k and m are the varied fitting parameters. These
parameters are derived from the linear relationship between cumulants and
wavelengths observed from Monte Carlo simulated TPSF. The values of these
unknown parameters are solved by means of fitting routine. Given here, this model is
used for determination of mean blood oxygen saturation. Hence, the absorbing
coefficient, aµ , is taken as the summation of product between concentration and
extinction coefficient of HbO2 and Hb. The values of the wavelength dependent
scattering coefficients are taken from the reports of Staveren et al. [51]. The
demonstration of the performance of this model is carried out using Monte Carlo
simulations to compare the measured A versus aµ relationship for different
scattering media in the blood medium.
2.3.3 Cubic function
A comparison between the non-linear light attenuation, A versus aµ relationship
obtained by Monte Carlo simulation and that given from MLB law has also been
extensively discussed by Kobayashi et al. [42] by means of fitting the attenuation
spectrum at each wavelength using the cubic function expressed as
3 2 2 3 2 2a a mel a mel mel a a mel mel a melA a b C c C dC e f C gC h iC jµ µ µ µ µ µ= + + + + + + + + +
(2.4)
where aµ denotes the absorption coefficient in the dermal layer and melC is the
concentration of melanin. The unknown parameters (i.e. aµ and melC ) and
coefficients a to j, with j as the offset, are numerically solved via a fitting procedure.
Four variables were considered during fitting, i.e. HbO2, Hb, melanin and the offset,
19
J. This function is derived using data obtained from the three-layered skin model
generated using Monte Carlo simulation The corresponding work compared the
results of obtained via Monte Carlo simulations and MLB law by deriving the
changes in absorbance, ΔZ between the four variables:
( ) ( ) ( ) ( ) ( ) ( )2
2
HbO Hb melaninHbO Hb melanin
Z Z ZZ J∂ ∂ ∂∆ = ∆ + ∆ + ∆ + ∆
∂ ∂ ∂ (2.5)
2.3.4 Power law model
The optical attributes of the skin tissue can be quantitatively assessed via a
mathematical model known as power law model [18]. This model explains a non-
linear relationship between the light attenuation of a diffused reflectivity, A, and aµ
in a semi-infinite medium,
( ) ( )0.351.06 1.45A µ µ= − (2.6)
Here, the symbol µ denotes the ratio of the absorbing coefficient and the
scattering coefficients given as a sµ µ . The absorption coefficient of the medium is
calculated by taking into account the volume fraction, V, of oxyhemoglobin (HbO2)
and deoxyhemoglobin (Hb),
( ) ( ) ( ) ( ) ( )2 22a HbO HbOa HbO a Hb(1 )V Vµ λ µ λ µ λ= + − (2.7)
The limitation to this model is, however, the approximation of tissue optical
properties is only confined within a specific spectral range of 45 10 0.1µ−× < < .
Hence, the applicability of this model is only valid for a sampling model using a red
or near infrared light source and a photon penetration depth of two millimeters.
2.3.5 Lambert − Beer law
Lambert − Beer law defines a linear relationship between light attenuation and
absorption in a non-scattering medium. The first part of the law, proposed by
20
Lambert, defines a linear relationship between light attenuation, A , and the light
pathlength, d . This law is complemented by Beer, who relates A with the medium's
absorption, aµ , to given the complete Lambert − Beer law in Equation 2.8 [27, 46].
aA dµ= (2.8)
Alternatively, changes in light attenuation in terms of the ratio of transmitted
intensity, I to incident intensity, 0I is given as [52]
0
log IAI
= (2.9)
Although there are uncertainties concerning the Lambert − Beer law due to
the inaccurate assumption of light pathlength which made it difficult to determine the
medium's absorption, the pertinence of this law has been vastly used as a basis in a
number of analytical model involving quantification of chromophores in an
absorbing medium [26, 53].
2.3.6 Modified Lambert Beer law
Modified Lambert Beer law proposed by Duling and Pittman [26] has been
extensively used for the assumption of percent blood oxygen saturation. This
quantification technique is based on the study of light absorption and scattering, with
spectral measurements taken at an isosbestic wavelength pair and a non-isosbestic
wavelength. The MLB equation is given as
( ) ( )aA G dλ µ λ= + (2.10)
where A represents the light attenuation and aµ is the wavelength dependent
absorption coefficient of absorber in the medium expressed in the unit mm-1.
Parameter G is the approximate attenuation offset due to scattering and d is taken as
the 'light pathlength'. This equation assumes a linear relationship between A versus
21
aµ of the absorbing medium. However, the medium's scattering coefficient, sµ ,
remains undeterred by the changes in the illuminating light wavelength.
Duling and Pittman [26] demonstrated the use of this model to determine
fractional concentration of hemoglobin derivatives (i.e. HbO2 and Hb) in the blood
medium given that sµ of two selected wavelengths are similar. Here, an isosbestic
wavelength pair (i.e. 520 nm and 546 nm shown in Figure 2.4) is used to solve for
the linear approximation in Equation 2.10, and a third non-isosbestic wavelength is
used to estimate the percent StO2 using the equation
( ) ( ) ( )
( )2 1 2 11 Hb HB 2 Hb Hb m Hb Hb
t 22 1 AB
S O N I I N I I
N
I I
I I
A A A
A A
ε ε ε ε ε ε
ε
− + − + −=
− (2.11)
where 1, 2I IA and NA are light attenuation values measured at two isosbestic
wavelengths and a non-isosbestic wavelength, respectively. In Equation 2.11,
subscripts I1 and I2 are repeatedly used to denote both the isosbestic wavelength
pair, while subscript N denotes the non-isosbestic wavelength. Meanwhile, ABNε
represents the differences between the extinction coefficient of oxyhemoglobin,
2HBOε , and extinction coefficient of deoxyhemoglobin, Hbε , at a non-isosbestic
wavelength expressed as N 2AB HBO HbN N
ε ε ε= − .
2.3.7 Extended Modified Lambert Beer model
There were drawbacks to the MLB law in which the StO2 value retrieved from the
present analytical model is less accurate, amongst which are mainly owing to the
poor assumption of light attenuation due to absorption and scattering process. To that
end, Huong and Ngu [13] proposed an attenuation model extended from the MLB
law, shown again in Equation 2.12, for improved estimation of percent
transcutaneous oxygen saturation.
( ) ( )0 a 0A G dλ µ λ= + (2.12)
22
Here, ( )aµ λ is defined as the sum of the product of concentration, C and
wavelength dependent extinction coefficient, ( )ε λ in Equation 2.13.
( ) ( ) ( )2 2a HbO HbO Hb HbC Cµ λ ε λ ε λ= + (2.13)
Considering blood HbO2 and Hb as the only light absorbers in the dermis
layer, the total hemoglobin concentration is given as 2HbO HbT C C= + . Given the
value of T, StO2 saturation is expressed in Equation 2.14 as
2HbOt 2S O
CT
= (2.14)
Substituting parameter T into Equation 2.13, the value of the total light
absorption, aµ is elaborated as
2 2 2 2a HbO HbO HbO HbO( ) ( )( )C T Cµ ε λ ε λ= + − (2.15)
where Equation 2.15 is further rearranged to give Equation 2.16.
( ) ( )( ) ( )( )2a HbO Hb t 2 HbS O Tµ ε λ ε λ ε λ= − + (2.16)
Huong and Ngu [13] proposed the EMLB model in Equation 2.17 as an
expression of a non-linear relationship between the light attenuation and medium's
absorption, thus, giving a more accurate StO2 value compared to that estimated using
the MLB law.
( )0 a 0 1 a 1( ) exp( )A G d G dλ µ λ λ λ µ= + + + − (2.17)
This extended equation is complemented by parameter 1G λ representing
light attenuation due to the wavelength dependent scattering process. The
exponential function in this model expressed light attenuation as a complex function
23
of dermal light scattering, assumed to change non-linearly with the 'light pathlength',
d1, and medium's absorption. The estimated value retrieved from the developed
analytic model shows a lower mean absolute error of 0.4% as compared to 10% by
MLB law, whilst light absorption by melanin is assumed linearly decreased with
wavelength. These results were validated using the attenuation data produced from
the Monte Carlo simulation code previously described by Chang et al. [54] . The
light absorption, aµ , used in this equation is based on the specific extinction
coefficient of oxyhemoglobin (2HbOε ) and deoxyhemoglobin ( Hbε ) in the wavelength
range of 520−600 nm given from the reports of Zijlstra et al. [46]. This is
considering the distinctive absorption of hemoglobin derivatives within the specific
wavelength as illustrated in Figure 2.7.
Figure 2.7: The wavelength dependent molar extinction coefficients of oxyhemoglobin (
2HbOε ), deoxyhemoglobin ( Hbε ) and carboxyhemoglobin ( COHbε ) compiled from the reports of Zijlstra et al. [46]
24
Experimental work employing EMLB model has revealed affirmative
outcome in continuous monitoring and measurement of percent mean blood
oxyhemoglobin (SmO2) and carboxyhemoglobin saturations (SmCO) as demonstrated
by Huong and Ngu [13]. The results were obtained via optical reflectance
spectroscopy performed on the left thumb of nine human subjects (five smoking and
four non-smoking individuals) and presented in Table 2.2. The result shows a
notably lower SmO2 and higher SmCO amongst smoking volunteers as compared to
that of non-smoking volunteers, a consequence of prolonged inhalation of carbon
monoxide (CO) or carboxyhemoglobin from cigarettes. Extensive studies on carbon
monoxide poisoning revealed CO has higher affinity of 240 times more likely to bind
with hemoglobin in blood, hence, restricting the delivery of blood carrying oxygen in
one's body [55].
Table 2.2: Percent mean blood oxyhemoglobin, SmO2 and carboxyhemoglobin, SmCO saturation in smoking and non-smoking individuals as reported by Huong and Ngu [56]
No. Volunteer group Reported mean blood oxygen saturation
SmO2 SmCO 1. Non-smokers 86.5 ± 1.6% 11.7 ± 1.6% 2. Smokers 81.9 ± 8.8% 16.2 ± 4.8%
Based on the reported results, this work has demonstrated the good
performance of the EMLB model for estimation of both blood saturation values via a
non-invasive approach. Unlike an average pulse oximeter, the employed EMLB
fitting model is able to provide SmO2 reading without referring to a look-up table for
estimation of blood saturation values. Likewise, the readings taken from a patient
suffering from cyanosis, a condition at which blood lacks oxygenated hemoglobin,
may result as incongruous given that the pulse oximeter detect mostly plasma in
blocked microcirculatory tissue [14]. In a situation of an individual with prolonged
exposure to carbon monoxide (CO), the application of a pulse oximeter often results
in a false-positive reading indicating 100% percent SaO2 value due to inaccuracy of
the oximeter in differentiating oxygenated hemoglobin (HbO2) and
carboxyhemoglobin (COHb) in blood [15]. Although an arterial blood gas analyzer is
able to provide accurate readings of both SmO2 and SmCO, this method has a
REFERENCES
[1] S. Schreml, R. Szeimies, L. Prantl, S. Karrer, M. Landthaler, and P. Babilas,
"Oxygen in acute and chronic wound healing," British Journal of
Dermatology, vol. 163, pp. 257-268, 2010.
[2] F. Gottrup, J. Apelqvist, and P. Price, "Outcomes in controlled and
comparative studies on non-healing wounds: recommendations," Journal of
Wound Care, vol. 19, p. 239, 2010.
[3] A. Zaini, "Where is Malaysia in the midst of the Asian epidemic of diabetes
mellitus?," Diabetes research and clinical practice, vol. 50, pp. S23-S28,
2000.
[4] B. Mustaffa, "Diabetes epidemic in Malaysia," Med J Malaysia, vol. 59, pp.
295-296, 2004.
[5] N. S. Raja, "Microbiology of diabetic foot infections in a teaching hospital in
Malaysia: a retrospective study of 194 cases," Journal of Microbiology
Immunology and Infection, vol. 40, p. 39, 2007.
[6] J. C. Phillips, "Understanding hyperbaric oxygen therapy and its use in the
treatment of compromised skin grafts and flaps," Plastic Surgical Nursing,
vol. 25, pp. 72-80, 2005.
[7] H. W. Hopf and M. D. Rollins, "Wounds: an overview of the role of oxygen,"
Antioxidants & Redox Signaling, vol. 9, pp. 1183-1192, 2007.
[8] J. Dissemond, K. Kröger, M. Storck, A. Risse, and P. Engels, "Topical
oxygen wound therapies for chronic wounds: a review," Journal of wound
care, vol. 24, pp. 53-63, 2015.
[9] G. S. Lazarus, D. M. Cooper, D. R. Knighton, D. J. Margolis, R. E. Pecoraro,
G. Rodeheaver, et al., "Definitions and guidelines for assessment of wounds
and evaluation of healing," Archives of Dermatology, vol. 130, pp. 489-493,
1994.
102
[10] L. C. Zhu, "Non-invasive optical technologies to monitor wound healing,"
Drexel University, 2007.
[11] L. Martinez, "A non-invasive spectral reflectance method for mapping blood
oxygen saturation in wounds," in Applied Imagery Pattern Recognition
Workshop, 2002. Proceedings. 31st, 2002, pp. 112-116.
[12] C. K. Sen, "Wound healing essentials: let there be oxygen," Wound Repair
and Regeneration, vol. 17, pp. 1-18, 2009.
[13] A. Huong and X. Ngu, "The application of extended modified Lambert Beer
model for measurement of blood carboxyhemoglobin and oxyhemoglobin
saturation," Journal of Innovative Optical Health Sciences, vol. 7, 2014.
[14] A. Al-Ali, "Patient monitor for determining microcirculation state," ed:
Google Patents, 2014.
[15] W. H. Maisel and R. J. Lewis, "Noninvasive measurement of
carboxyhemoglobin: how accurate is accurate enough?," Annals of
emergency medicine, vol. 56, pp. 389-391, 2010.
[16] A. Huong and X. Ngu, "The application of extended modified Lambert Beer
model for measurement of blood carboxyhemoglobin and oxyhemoglobin
saturation," Journal of Innovative Optical Health Sciences, 2013.
[17] R. R. Anderson and J. A. Parrish, "The optics of human skin," Journal of
Investigative Dermatology, vol. 77, pp. 13-19, 1981.
[18] A. Vogel, V. V. Chernomordik, S. G. Demos, R. Pursley, R. F. Little, Y. Tao,
et al., "Using noninvasive multispectral imaging to quantitatively assess
tissue vasculature," Journal of Biomedical Optics, vol. 12, pp. 051604-
051604-13, 2007.
[19] A. Pifferi, P. Taroni, G. Valentini, and S. Andersson-Engels, "Real-time
method for fitting time-resolved reflectance and transmittance measurements
with a Monte Carlo model," Applied Optics, vol. 37, pp. 2774-2780, 1998.
[20] T. Durduran, R. Choe, W. Baker, and A. Yodh, "Diffuse optics for tissue
monitoring and tomography," Reports on Progress in Physics, vol. 73, p.
076701, 2010.
[21] W. B. Baker, A. B. Parthasarathy, D. R. Busch, R. C. Mesquita, J. H.
Greenberg, and A. Yodh, "Modified Beer-Lambert law for blood flow,"
Biomedical Optics Express, vol. 5, pp. 4053-4075, 2014.
103
[22] Y.-W. Ri, S.-H. Jong, and S.-J. Im, "Theoretical prediction of the source-
detector separation distance suited to the application of the spatially resolved
spectroscopy from the near-infrared attenuation data cube of tissues," arXiv
preprint arXiv:1409.4246, 2014.
[23] J. Murkin and M. Arango, "Near-infrared spectroscopy as an index of brain
and tissue oxygenation," British Journal of Anaesthesia, vol. 103, pp. i3-i13,
2009.
[24] H.-Y. Tsai, K.-C. Huang, H.-C. Chang, and C.-H. Chang, "A study on oxygen
saturation images constructed from the skin tissue of human hand," in
Instrumentation and Measurement Technology Conference (I2MTC), 2013
IEEE International, 2013, pp. 58-62.
[25] M. H. Smith, "Optimum wavelength combinations for retinal vessel
oximetry," Applied optics, vol. 38, pp. 258-267, 1999.
[26] R. N. Pittman and B. R. Duling, "A new method for the measurement of
percent oxyhemoglobin," Journal of applied physiology, vol. 38, pp. 315-320,
1975.
[27] A. K. C. Huong, "Spectroscopic analysis of scattering media via different
quantification techniques," Thesis, University of Nottingham, 2012.
[28] F. C. Delori, "Noninvasive technique for oximetry of blood in retinal
vessels," Applied optics, vol. 27, pp. 1113-1125, 1988.
[29] L. Caspary, J. Thum, A. Creutzig, D. Lubbers, and K. Alexander,
"Quantitative reflection spectrophotometry: spatial and temporal variation of
Hb oxygenation in human skin," International Journal of Microcirculation,
vol. 15, pp. 131-136, 1995.
[30] H. Arimoto and H. Furukawa, "Retinal blood oxygen saturation mapping by
multispectral imaging and morphological angiography," in Engineering in
Medicine and Biology Society, 2007. EMBS 2007. 29th Annual International
Conference of the IEEE, 2007, pp. 1627-1630.
[31] J. A. Freeberg, J. Benedet, C. MacAulay, L. A. West, and M. Follen, "The
performance of fluorescence and reflectance spectroscopy for the in vivo
diagnosis of cervical neoplasia; point probe versus multispectral approaches,"
Gynecologic oncology, vol. 107, pp. S248-S255, 2007.
[32] J. E. Freeman and E. N. Lewis, "Multispectral/hyperspectral medical
instrument," ed: Google Patents, 2013.
104
[33] H. Akbari and Y. Kosugi, "Hyperspectral imaging: A new modality in
surgery," Recent Advances in Biomedical Engineering, ed. GR Naik, pp. 223-
240, 2009.
[34] G. Bianco, F. Bruno, and M. Muzzupappa, "Multispectral data cube
acquisition of aligned images for document analysis by means of a filter-
wheel camera provided with focus control," Journal of Cultural Heritage,
vol. 14, pp. 190-200, 2013.
[35] M. Aikio, Hyperspectral prism-grating-prism imaging spectrograph:
Technical Research Centre of Finland Heinola, Finland, 2001.
[36] T. Binzoni, A. Vogel, A. Gandjbakhche, and R. Marchesini, "Detection limits
of multi-spectral optical imaging under the skin surface," Physics in Medicine
and Biology, vol. 53, p. 617, 2008.
[37] B. M. Carlson, Human Embryology and Developmental Biology: with
STUDENT CONSULT Online Access: Elsevier Health Sciences, 2012.
[38] B. Amirlak, L. Shahabi, S. Javaheri, F. Talavera, and W. Stadelmann, "Skin
anatomy," www. emedicinemedscapecom/article/1294744-overview, 2011.
[39] I. V. Meglinski and S. J. Matcher, "Quantitative assessment of skin layers
absorption and skin reflectance spectra simulation in the visible and near-
infrared spectral regions," Physiological Measurement, vol. 23, p. 741, 2002.
[40] R. Zhang, W. Verkruysse, B. Choi, J. A. Viator, B. Jung, L. O. Svaasand, et
al., "Determination of human skin optical properties from spectrophotometric
measurements based on optimization by genetic algorithms," Journal of
Biomedical Optics, vol. 10, p. 024030, 2005.
[41] V. Barun, A. Ivanov, A. Volotovskaya, and V. Ulashchik, "Absorption
spectra and light penetration depth of normal and pathologically altered
human skin," Journal of applied spectroscopy, vol. 74, pp. 430-439, 2007.
[42] M. Kobayashi, Y. Ito, N. Sakauchi, I. Oda, I. Konishi, and Y. Tsunazawa,
"Analysis of nonlinear relation for skin hemoglobin imaging," Optics
Express, vol. 9, pp. 802-812, 2001.
[43] T. Maeda, N. Arakawa, M. Takahashi, and Y. Aizu, "Monte Carlo simulation
of spectral reflectance using a multilayered skin tissue model," Optical
Review, vol. 17, pp. 223-229, 2010.
[44] A. F. M. Hani, R. Baba, N. Shamsuddin, and H. Nugroho, "Determination of
melanin types and relative concentrations: an observational study using a
105
non‐invasive inverse skin reflectance analysis," International Journal of
Cosmetic Science, vol. 36, pp. 451-458, 2014.
[45] D. Yudovsky and L. Pilon, "Rapid and accurate estimation of blood
saturation, melanin content, and epidermis thickness from spectral diffuse
reflectance," Applied Optics, vol. 49, pp. 1707-1719, 2010.
[46] W. G. Zijlstra, A. Buursma, and O. W. van Assendelft, Visible and near
infrared absorption spectra of human and animal haemoglobin:
determination and application: VSP, 2000.
[47] D. Parker, "Optical device," ed: Google Patents, 2007.
[48] J. D. Atencio, S. Jacques, and S. V. y Montiel, "Monte Carlo Modeling of
Light Propagation in Neonatal Skin."
[49] W.-F. Cheong, S. A. Prahl, and A. J. Welch, "A review of the optical
properties of biological tissues," IEEE journal of quantum electronics, vol.
26, pp. 2166-2185, 1990.
[50] A. Sassaroli, F. Martelli, and S. Fantini, "Perturbation theory for the diffusion
equation by use of the moments of the generalized temporal point-spread
function. I. Theory," JOSA A, vol. 23, pp. 2105-2118, 2006.
[51] H. J. Van Staveren, C. J. Moes, J. van Marie, S. A. Prahl, and M. J. Van
Gemert, "Light scattering in lntralipid-10% in the wavelength range of 400–
1100 nm," Applied optics, vol. 30, pp. 4507-4514, 1991.
[52] I. Alabboud, "Human retinal oximetry using hyperspectral imaging," Heriot-
Watt University, 2009.
[53] A. K. Huong and X. T. Ngu, "In situ monitoring of mean blood oxygen
saturation using Extended Modified Lambert Beer model," Biomedical
Engineering: Applications, Basis and Communications, vol. 27, p. 1550004,
2015.
[54] P. Chang, J. Walker, K. Hopcraft, B. Ablitt, and E. Jakeman, "Polarization
discrimination for active imaging in scattering media," Optics
communications, vol. 159, pp. 1-6, 1999.
[55] N. Pavelchak, E. Franko, J. Zhu, and R. DePersis, "Assessment of carbon
monoxide exposure during the operation of indoor drive-through mass
vaccination clinics," Disaster medicine and public health preparedness, vol.
3, pp. 158-162, 2009.
106
[56] A. Huong and X. Ngu, "Noninvasive diagnosis of carbon monoxide
poisoning using Extended Modified Lambert Beer Model," in 2nd
International Conference on Electronic Design (ICED), 2014, 2014, pp. 265-
269.
[57] E. S. Papazoglou, M. S. Weingarten, L. Zubkov, L. Zhu, S. Tyagi, and K.
Pourrezaei, "Optical properties of wounds: diabetic versus healthy tissue,"
Biomedical Engineering, IEEE Transactions on, vol. 53, pp. 1047-1055,
2006.
[58] L. Braiman-Wiksman, I. Solomonik, R. Spira, and T. Tennenbaum, "Novel
insights into wound healing sequence of events," Toxicologic Pathology, vol.
35, pp. 767-779, 2007.
[59] S. D. Bateman, "Topical haemoglobin spray for diabetic foot ulceration,"
British Journal of Nursing, vol. 24, 2015.
[60] R. Norris, "A topical haemoglobin spray for oxygenating chronic venous leg
ulcers: a pilot study," British Journal of Nursing, vol. 23, pp. S48-S53, 2014.
[61] D. A. Scherson and M. I. Burk, "Oxygen producing bandage," ed: Google
Patents, 1999.
[62] R. Ewall, "Multi-purpose wound dressing," ed: Google Patents, 1997.
[63] L. Khaodhiar, T. Dinh, K. T. Schomacker, S. V. Panasyuk, J. E. Freeman, R.
Lew, et al., "The use of medical hyperspectral technology to evaluate
microcirculatory changes in diabetic foot ulcers and to predict clinical
outcomes," Diabetes Care, vol. 30, pp. 903-910, 2007.
[64] S. Zhang, S. Gnyawali, J. Huang, W. Ren, G. Gordillo, C. K. Sen, et al.,
"Multimodal imaging of cutaneous wound tissue," Journal of biomedical
optics, vol. 20, pp. 016016-016016, 2015.
[65] D. Nouri, Y. Lucas, S. Treuillet, R. Jolivot, and F. Marzani, "Colour and
multispectral imaging for wound healing evaluation in the context of a
comparative preclinical study," in SPIE Medical Imaging, 2013, pp. 866923-
866923-10.
[66] I. Djermanov, S. Iordanova, and I. Koleva, "Spectrograph coupled with CCD
module for high resolution spectroscopy measurements," in Journal of
Physics: Conference Series, 2007, p. 012029.
[67] C. E. Thorn, S. J. Matcher, I. V. Meglinski, and A. C. Shore, "Is mean blood
saturation a useful marker of tissue oxygenation?," American Journal of
107
Physiology-Heart and Circulatory Physiology, vol. 296, pp. H1289-H1295,
2009.
[68] R. Widenhorn, M. M. Blouke, A. Weber, A. Rest, and E. Bodegom,
"Temperature dependence of dark current in a CCD," in Electronic Imaging
2002, 2002, pp. 193-201.
[69] A. A. Kamshilin, V. Teplov, E. Nippolainen, S. Miridonov, and R.
Giniatullin, "Variability of microcirculation detected by blood pulsation
imaging," PloS one, vol. 8, p. e57117, 2013.
[70] J. E. Sinex, "Pulse oximetry: principles and limitations," The American
journal of emergency medicine, vol. 17, pp. 59-66, 1999.
[71] I. Ghoneimi and R. Bang, "Use of Solcoseryl in minor burns," Annals of
Burns and Fire Disasters, vol. 10, pp. 219-222, 1997.
[72] F. Werdin, M. Tennenhaus, H.-E. Schaller, and H.-O. Rennekampff,
"Evidence-based management strategies for treatment of chronic wounds,"
Eplasty, vol. 9, 2009.
[73] C. Fife, D. Smart, P. Sheffield, H. Hopf, G. Hawkins, and D. Clarke,
"Transcutaneous Oximetry in Clinical Practice: Consensus statements from
an expert panel based on evidence.*," Undersea & Hyperbaric Medicine, vol.
36, p. 43, 2009.
[74] D. Gray, R. White, P. Cooper, and A. Kingsley, "Applied wound
management and using the wound healing continuum in practice," Wound
Essentials, vol. 5, p. 39, 2010.
[75] C. Dowsett and H. Newton, "Wound bed preparation: TIME in practice,"
WOUNDS UK, vol. 1, p. 58, 2005.
[76] A. Brown, "The role of debridement in the healing process," Nursing times,
vol. 109, pp. 16-19, 2012.